Abstract: Urbanization is arguably the most dramatic form of land transformation with profound impacts on the natural environment, biological diversity and human life. Human activities have inevitably altered the structure, function, and dynamics of ecological systems, such as fragmentation of natural elements in the urban landscape. Therefore, it has become increasingly important for large-scale ecological research and applications (e.g., urban landscape planning, land use planning, and biodiversity conservation) to consider the ecological consequences of these dramatic land transformations. It is a major task for urban landscape planners to construct effective and harmonious urban greenbelt networks and maintain a sustainable urban development environment. Based on a land use map created from a Landsat-TM satellite image from the year 2000, a map of a greenbelt system from 2002 and a map of a planned greenbelt system, we used landscape spatial pattern analysis and a series of landscape pattern metrics on a GIS platform to assess the present situation and the planned greenbelt system for Xiamen Island. Based on these analyses, a number of new planning scenarios were designed by using network analysis methods for optimizing the ecological network of Xiamen Island. Some indices that reflected corridor characteristics, such as corridor length and corridor density, were also calculated for every planning scenario. Using network indices such as α、β、γ indices, cost ratio and corridor metrics, an optimal planning scenario E was selected among those planning scenarios of ecological networks. The optimal planning scenario E was then overlaid on the existing plans for a greenbelt system on a GIS platform. A series of landscape metrics were then calculated to assess the improvement of the optimal planning scenario E on the ecological network of the existing greenbelt system plans for Xiamen Island. Our results showed that the greenbelt system plan for Xiamen Island was an improvement as indicated by a decrease in patch density and increase in edge density, mean patch fractal dimension on the patch level, landscape diversity, landscape evenness, and landscape connectivity. Compared to the existing greenbelt system plan, the optimal planning scenario E would improve the network connectivity and network circuitry considerably. The optimal planning scenario E greenbelt system could improve the degree of landscape fragmentation, increase the shape complexity of greenbelt patches and increase the landscape connectivity greatly. Our results indicated that the methods used were able to integrate landscape pattern metrics with network analyses and quantitatively assess the present situation and the rationality of planning for urban greenbelt systems. Also we demonstrated that these methods could be used to optimize planning scenarios for urban ecological networks to make them more aligned with the principles of conformity, harmony, circulation, safety, diversity and sustainability.